It has been known from time immemorial that fingerprints can be used to identify people. Initially fingerprints were taken manually by inking the first joint of a finger and pressing this onto a sheet of paper. A pattern of lines was thus printed onto the sheet of paper, where the lines corresponded to the ridges and the spaces between the lines corresponded to the grooves in the skin of the finger. Now fingerprints are taken using sensors that detect the pattern on the skin on the part of the finger that is held against the sensor.
Traditionally, fingerprint technology has been used primarily for the identification of criminals, but in recent years it has also begun to be used for access control.
Known access control systems are based on a person who is to be given access to something, recording his fingerprint under secure conditions. The recorded fingerprint is stored in a memory, for example a database or on a personal data carrier.
When the access control is to be carried out, the person places his finger on a sensor which records a current fingerprint. This recorded current fingerprint is compared with the previously recorded fingerprint or fingerprints that are in the memory, in order to check whether the recorded current fingerprint is the same as this previously recorded fingerprint or any one of these previously recorded fingerprints and accordingly belongs to an approved person. If so, the system sends a signal indicating that such is the case. The signal can control the access to, for example, premises, a computer or information, as applicable.
Checking whether a current recorded fingerprint is the same as a previously recorded fingerprint is carried out using hardware or software and on the basis of digital representations of the fingerprints. The fingerprint must be represented at high resolution, as only small details distinguish different fingerprints from each other. This means that there is a large amount of information that must be stored for each fingerprint, and also that there is a lot of information that has to be compared in order to determine whether two fingerprints originate from the same person.
One way of reducing the amount of information that is stored and compared is to utilise “feature extraction”. According to this, the fingerprint is recorded from the person who is to be granted access to whatever is protected by the system, special features in the fingerprint are sought, such as points where the ridges divide and points where the ridges end, and the positions of these features are determined. Only the position information and the categorisation of the features is stored in the system's memory. When a person's authorisation is to be checked this person's fingerprint is recorded again and then a check is made whether the same features recur in the same places in the current recorded fingerprint as in the previously stored fingerprint.
The sensors that are used in fingerprint systems are based on various detection techniques. There are optical sensors, which essentially comprise a transparent surface, against which the user holds his finger, a lens system and the sensor proper which has a light-sensitive surface. The lens system projects a reduced image of the fingerprint onto the sensor. Optical sensors have the disadvantage of being unwieldy in design, on account or the lens system.
There are also so-called silicon sensors, where the user's finger is in direct contact with the active surface of the sensor. At present the most common silicon sensors are capacitive. They have a sensor surface of at least 100 mm2. The capacitive silicon sensors are relatively expensive as they require a large silicon area.
In addition, there is provided so called swipe sensors. When a fingerprint is to be read, the user passes his finger over the sensor which records, at a high frequency and using e.g. capacitive detection, “images” of the part of the finger that is over the sensor at that moment. The “images” are thereafter put together into a single “image” of the whole fingerprint. The whole image is thereafter compared in the normal way with a previously recorded image. This sensor requires a smaller silicon area than the sensors that record a whole fingerprint from a finger that is held still.
A problem with swipe sensors is to efficiently use the plurality of images acquired by the sensor for checking of the fingerprint.